This model uses several easily ascertained items plus bone mineral density (BMD) to calculate 10-year absolute fracture risk. The model begins by assigning the average fracture rate based on the patient’s age. These population rates come from a publication that used US hospital discharge data for 2006. 1The model calculates differences between the patient and the reference population in several clinical risk factors including bone mineral density (BMD). Each risk factor’s relative risk is used to modify the expected fracture rate. The risk factors and relative risks have been closely aligned to those used in FRAX™ 2.
Excellent risk-assessment models 3,4 exist for other diseases and have been translated into computer software that calculates absolute risk of event from multiple variables. These models include the one widely used to predict 5- and 10-year risks for breast cancer 3 and the National Cholesterol Education Program’s model for predicting 10-year risk of coronary heart disease.4 Both models have been used in developing best practice guidelines, are widely recommended to guide therapeutic intervention, and are well accepted by clinicians. Cost-effectiveness analyses 5 and treatment recommendations based on these analyses 6 have now been published for osteoporosis drugs.
The WHO FRAX™ model essentially replaces all previous web-based models because it has been extensively validated.
The ABH FRC model and FRAX™ model use the same input variables, the same base fracture rates, and the same relative risks (multipliers). 2 Both can be applied to men and 4 different ethnicities. Both output the 10-year risk of hip and any one of 4 fractures (hip, wrist, humerus, clinical spine). One major difference is that the ABH model provides a graphic display of risk, comparing it to the expected risk, and categorizing it as low, medium, or high. A second difference is the FRAX™ model performs more adjustments for interactions between variables. Thus, users of these models should get very similar but not identical results.
The ABH and FRAX™ models provide 10-year absolute fracture risk for a hip fracture and for any one of 4 fractures (hip, spine, humerus, or wrist). The mix of fractures changes with age: at 45-49 years, hip fractures constitute only 7% but increase to 50% at 75 years.7 In contrast, the proportion of wrist fractures decreases with age, from 60% among the younger group to 26% among the older.
Both the ABH and FRAX™ models include four clinical osteoporotic fracture locations (i.e., spine, wrist, humerus, and hip) for several reasons: low BMD is a demonstrated risk factor for these fracture types;7 these fractures constitute the majority of osteoporotic (fragility-type) fractures;8 accurate rates of fracture incidence for the age range of interest were available from reliable published data1 and these fractures are costly to treat and can seriously alter quality of life.9
Both the ABH and FRAX™ models aim to keep the model data entry simple and in an easily understood format. Both use risk factors whose individual contributions to fracture risk have been estimated in large epidemiologic studies using multivariable models; the risk factors are the same as those used in cost-effectiveness analyses.5
The WHO model is the standard in the field and has been incorporated into BMD reporting. In the ABH model, we have tried, as much as possible, to use the rates and relative risks that appear in the WHO model. 2 The ABH model adds graphic display and categorization of risk—that should be helpful in discussing prevention and treatment strategies and risks and benefits of treatment.
Models could promote awareness of the need for treatment among overlooked women in their 60s who have a number of risk factors but no clinically apparent osteoporosis (i.e. T-score not <-2.5). Fracture models appear to have the greatest value in solving the common clinical conundrum of low bone density (osteopenia) in healthy, early postmenopausal women. Giving a patient her absolute risk and discussing risk reduction expected from treatment in an easily understandable context enriches the quality and accuracy of information both health care providers and patients use for making decisions.
Both ABH and FRAX™ models can be used for men and for 4 different ethnicities (Caucasian, Hispanic, Black, Asian). Simply select gender and race/ethnicity in the input fields.
This model is based on fracture rates in untreated women and does not account for osteoporosis treatment effects. Estimates of the fracture risk reduction from regular long-term bisphosphonate therapy are in the order of 25-35%. 5,6,10 Leslie et al 11, on the basis of a large population study, concluded that “the FRAX tool can be used to predict fracture probability in women currently or previously treated for osteoporosis”. In that study, bisphosphonate use did not substantially change the categorization of women in the population to low, medium, or high risk.
This model does not capture the additional risks imposed by falling and frailty because there are not studies that provide a relative risk for fractures based on these variables.
Cost-effectiveness analyses and treatment guidelines based on US costs and fracture rates have been published.5,6 The thresholds cited are 3% or more for 10-year risk of hip fracture and 20% or more for 10-year risk of any one of 4 fractures. The ABH model shows rates above the 20% threshold as high risk, includes an intermediate zone (10-20%), as well as a low risk zone (<10%). These same percentage cut points are used for cardiovascular disease (National Cholesterol Education Program)4 and have been adopted by expert osteoporosis groups.12 In counseling patients about the risks and benefits of treatments for osteoporosis, we find that most individuals start to worry about their risk when it reaches “double digits” and 20% (1 in 5) is usually of great concern. Providing absolute risks rather than relative risks allows patients to make better informed decisions about the risks of fracture and the benefits of therapy.13